2. Securitisation in Russia and the CIS: Moody’s Perspective Daniel Mumzhiu, Analyst – Business Development, Structured Finance Group June 21, 2007

3. 2 Agenda What is Moody’s looking for from CIS issuers? Obstacles/Solutions to Securitisation in CIS How High Can the Rating Be? RMBS Rating Methodology in Brief

4. 3 What is Moody’s Looking for From CIS Issuers?

5. 4 Data Requirements Accurate and complete data is key for analysis so prepare EARLY As much historical data as possible – Static pool data preferred Check data requirements in advance – Data requirements depend on asset class and may be different than in Western Europe Data should be in the correct format Third party audit to 99/1 confidence level required

6. 5 Originator Operational Review Company and management overview Origination and underwriting standards Quality Control Risk Management Servicing and collection process Pool/portfolio characteristics and performance Other documents

7. 6 Management Staff experience, training, compensation Loan Administration Arrears management Loss mitigation Asset management IT systems and reporting General Quality and guidelines (including history, structure and strategy) Financial stability Servicer Review – Nine Areas

8. 7 Obstacles and Solutions to Securitisation in CIS

9. 8 Obstacles to Securitization in the CIS Regulatory issues Legal issues – Do not preclude Moody’s from assigning IG ratings to transactions from originators with low-ratings (assuming the transaction is properly structured) Data collection and history Unrated originators and servicers Depth of local capital markets FX Convertibility and moratorium risk

10. 9 Suggestions – Early Preparation For a Transaction Discuss legal issues with lawyers & rating agency Decide and discuss rating target – See what enhancement/structure may be needed Prepare portfolio data, including data format Prepare historical performance data Consider back-up servicer Determine structure – External support? (PRI, wrap, guarantee)

11. 10 How High Can the Rating Be?

12. 11 Risk Layers and Rating Scales Assets Origination Serivicng History Structure Systemic risks Political risk Currency swap IR swap Liquidity Back-up servicing ------------- LC- National Scale (NSR) Legal Enforcement Fraud risk Payment systems Quality of data/IT Overall stability ------------ LC- Global Scale Transferability Convertibility Expropriation -------------------- FX- Global Scale

13. 12 Piercing the Foreign Currency Ceiling – How? Foreign currency ceiling may be pierced based on: (1) highly rated local currency obligation combined with (2) liquidity facility or political risk insurance Country’s foreign currency ceiling (Example: A2 for Russia) Country’s local currency ceiling (LCC) (Example: A1 for Russia) Aaa A rating can exceed the LCC only by an external guarantee or insurance “wrap” Assets Credit Quality Credit Enhancement: subordination, reserve, excess spread SYSTEMICRISKSSYSTEMICRISKS LEGALRISKSLEGALRISKS

14. 13 Foreign and Local Currency Ceilings CountryForeign Currency CeilingLocal Currency Ceiling BulgariaA1Aa3 CroatiaA1Aa1 Czech RepublicAa1Aaa EstoniaAa1Aaa HungaryAa1Aaa KazakhstanA2A1 LatviaAa1Aaa LithuaniaAa1Aaa PolandAa1Aaa RomaniaA1Aa3 RussiaA2A1 SlovakiaAa1Aaa SloveniaAaa TurkeyBa1A2 UkraineBa3A3

15. 14 RMBS Rating Methodology in Brief

16. 15 Moody’s Rating – Expected Loss n Ratings measure credit risk n Probability of a default n Severity of a loss Expected loss = Probability of default x Severity Example: – Prob. Def. = 5% – Severity of Loss = 20% (recovery rate = 80%) – EL = 5% x 20% = 1% – And, if the average life of the security is 6 years…

17. 16 Moody’s Idealised Expected Loss Table

18. 17 Moody’s RMBS Analysis - Overview Main tool for quantitative RMBS analysis – Cash Flow Model (MARCO) to determine expected loss for each class of transaction Major input factor in Cash Flow Model – Loss Distribution Determination of Loss Distribution – Historical loss data for estimation of expected loss and volatility – Usage of scoring models (MILAN) for estimation of volatility and/or expected loss

19. 18 MILAN –Standardised Scoring Model MILAN is the result of an in depth analysis and comparison of major EMEA RMBS markets Although being a standardised model, each MILAN version addresses specific features for each country MILAN is a Scoring Model, which scores each single loan compared to a benchmark loan and the total portfolio to a benchmark portfolio Defines “Adjusted MILAN CE“ the committeed credit enhancement level that is necessary in a simple pass-through transaction to get to target rating

20. 19 Overview of MILAN Step 1 Definition of benchmark loan and portfolio Step 1 Definition of benchmark loan and portfolio Step 3 Loss severity Step 3 Loss severity Step 2 Default frequency Step 2 Default frequency Step 4 Definition of benchmark credit enhancement subject to minimum credit enhancement Step 4 Definition of benchmark credit enhancement subject to minimum credit enhancement MILAN at a Glance - Steps 1 to 4

21. 20 MILAN at a Glance (2) Step 5-8 Each loan level adjustments Step 5-8 Each loan level adjustments Step 9 Each individual loan Aaa CE Step 9 Each individual loan Aaa CE Step 10 Portfolio level adjustments Step 10 Portfolio level adjustments Step 11/12 MILAN Aaa CE Adjusted MILAN Aaa CE after committee Step 11/12 MILAN Aaa CE Adjusted MILAN Aaa CE after committee MILAN at a Glance - Steps 5 to 12

22. 21 Benchmark Credit Enhancement Benchmark credit enhancement (“CE Bench”) is the expected loss of a benchmark loan in a recessive scenario Definition and Basis Default frequency based on: Loan-to-Value ratio (LTV) Loss severity based on: costs of foreclosure time to foreclosure missed interest minimum CE house price stress Two parts

23. 22 LTV/Default Curve Example LTV 60 % FD 4.7% LTV 80% FD 10.9 % LTV 100 % FD 17.3 % LTV Frequency of Default

24. 23 Loss severity Property Market Value Stressed Property Value Foreclosure Costs Prior Ranks + Missed Interest Current Loan Balance Loss Severity House Price Stress Rate

25. 24 MILAN Aaa CE – Example CE Bench7.06% 1. Property Adjustments3.53% 2. Loan Adjustments1.06% 3. Borrower Adjustments3.53% 4. Performance Adjust.0.00% 5. Originator/Servicer0.76% Min Aaa CE4.00% MILAN Aaa CE each loan15.94% 6. Portfolio Adjustments1.40% MILAN Aaa CE for Portfolio17.34% Adjusted MILAN Aaa CE17.50%

26. 25 Lognormal loss distribution used in MARCO

27. 26 Calculation of ratings Once the lognormal curve has been built, the MARCO model will run as follows: The model will run a number of loss scenarios, computed with the lognormal curve, on the portfolio and the waterfall The loss rate and the average life on each tranche of notes resulting from each loss scenario on the portfolio and from the deal structure will be computed and will be weighted by the corresponding probabilities of the scenarios The result will be the expected loss and the weighted average life for each tranche, which are used to derive ratings from the idealised expected loss tables

28. 27 www.moodys.com Contact: daniel.mumzhiu@moodys.com © Copyright 2006, Moody’s Investors Service, Inc. and/or its licensors including Moody’s Assurance Company, Inc. (together, “MOODY’S”). All rights reserved. ALL INFORMATION CONTAINED HEREIN IS PROTECTED BY COPYRIGHT LAW AND NONE OF SUCH INFORMATION MAY BE COPIED OR OTHERWISE REPRODUCED, REPACKAGED, FURTHER TRANSMITTED, TRANSFERRED, DISSEMINATED, REDISTRIBUTED OR RESOLD, OR STORED FOR SUBSEQUENT USE FOR ANY SUCH PURPOSE, IN WHOLE OR IN PART, IN ANY FORM OR MANNER OR BY ANY MEANS WHATSOEVER, BY ANY PERSON WITHOUT MOODY’S PRIOR WRITTEN CONSENT.